DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Election/Restrictions
In response to the Restriction dated December 23, 2025, the Applicant has elected without traverse group I, which are claims 1-14, in the reply filed on January 30, 2026. The applicant presently cancels claims 15-21. The applicant has added new claims 22-28 in the claims filed January 30, 2026. Therefore, claims 1-14 and 22-28 are pending and subject to examination.
In response to the “Informalities” section of the Restriction dated December 23, 2025, the applicant has also made minor amendments to claims 1 and 8. The examiner accepts these amendments and agrees that these claims are now free of such informalities.
Claim Objections
Claims 3, 10, 22, and 24 recite an “AI based system”. This should read “AI-based system. For examination purposes, that is how it will be interpreted. Appropriate response or correction is respectfully required.
Claim Rejections - 35 USC § 102
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1-14 and 22-28 are rejected under 35 U.S.C. 102(a)(2) as being anticipated by Yan et al. (US2025/0200979).
Regarding claim 1, Yan discloses:
A method comprising:
determining, from a plurality of physical object observations derived from sensor-acquired data and non-sensor-acquired data collected for a vehicle, a plurality of input feature vectors in a latent space of three dimensions (see Yan paragraph 0014 for the sensors on the host vehicle. See Fig. 2 for sensing data 210, traffic lights 220, and map information 124. See paragraph 0032 for additional sensors. See paragraph 0052 for “feature vectors” in “different dimensions”. See paragraph 0054 for the “feature vector” including in “a multi-dimensional embedding space”. See paragraph 0058 for tokens in three dimensions.),
wherein the three dimensions include
a time dimension (see paragraph 0053 for a “temporal dimension”),
a feature dimension and (the present published disclosure, Shamsoshoara (US2025/0196892) uses the word “feature” extremely broadly. Paragraph 0125 refers to “different feature types (e.g., different vehicles/objects)”. With that in mind, see Yan, paragraph 0052 for “feature vectors” such as roadgraph information having a dimension Drx1.)
an embedding size dimension (in the present published disclosure see paragraph 0101 for the “The embeddings or vectors may be of an equal length (or size/dimension) referred to as ‘embedding size’”. With that in mind, see Yan paragraph 0052 for “embeddings of a common dimension”.);
applying a plurality of attention heads implemented in one or more transformer networks of an artificial intelligence (AI)-based system to the input feature vectors in the latent space to generate attention scores forming a plurality of attention layers (in the present published disclosure, see Fig. 4, step. 404. What is an “attention head” in the disclosure? Paragraph 0067 recites that “multi-head attention (networks/blocks) may include…a plurality of attention heads each of which represents a query (Q), key (K) and value (V) (attention subnetwork as illustrated in Fig. 2C.” Paragraph 0068 recites that “Through these attention heads of QKV (attention) subnetworks (or sub-blocks), the multi-head attention (networks/blocks) in the attention-based model 204 or the transformer (neural network) in the AI tactical planner 102 can process input vectors or embeddings represented in a late space 208 of Fig. 2C.” Paragraph 0070 refers to “the attention heads of QKV subnetworks can operate in parallel in real-time to pay attention across different features represented in the latent space 208”. Paragraph 0065 teaches the “attention-based model 204” and paragraph 0066 teaches that the “multi-head (self) attention networks (blocks)…may be implemented as a…part of the attention-based model 204”, which can be seen in Fig. 2B. Paragraphs 0120-0121 teach “attention scores”. Paragraph 0127 teaches that “the attentions paid or attention scores computed” in model 204 can be switched or swapped between the two dimensions of feature and time in the latent space 208. This way “the AI-based models as described herein are trained or applied to respond to a wide variety of contextual, temporal or latent relations among or between values or features alongside the time axis/dimension as well as the features axis-dimension in the late space 208.”
With that in mind, see Yan paragraph 0056 for a neural network with an “attention-based transformed architecture” including spatial and temporal encoder blocks. See paragraph 0057 for multi-axes attention blocks that “compute attention scores”. Each attention block can including “one or more multi-head self-attention layers”. These blocks can be stacked together. See also Fig. 4);
generating, based at least in part on the attention scores in the plurality of attention layers, one or more target predictions relating to navigation operations of the vehicle (see paragraph 0033 for the system being configured for “selecting a particular path through the immediate driving environment, which can include selecting a traffic lane, negotiating a traffic congestion…and so on.” See Fig. 4 and paragraph 0071 for “predicted trajectories 442”.);
interacting with a driver assistant sub-system of the vehicle to cause a vehicle propulsion operation to be performed in accordance with the one or more target predictions (in the present disclosure see paragraph 0077 for the host vehicle having a lane change assistant (LCA) among other ADAS functions. See paragraph 0056 for the system potentially using “lane data and or object data represent[ing] history information of all vehicles and/or objects in the same scene or on the same road section traveled by the vehicle/ego up to the latest time or the current wall clock time.” See paragraph 0041 for the AT-based “target predictions” including lane change commands. See paragraphs 0043-0044 for predicting future lane changes and “passing the future lane change predictions…to the travel assistant to…perform specific lane changes as predicted.”
With that in mind, see Yan paragraph 0033 for the system being configured for “selecting a particular path through the immediate driving environment, which can include selecting a traffic lane, negotiating a traffic congestion…and so on.” See also paragraph 0017.).
Regarding claim 2, Yan discloses the method of Claim 1.
Yan further discloses:
The method of Claim 1, wherein
the plurality of attention heads includes
a first attention head configured to apply first self-attention to the plurality of input feature vectors across the time dimension of the latent space (see Yan paragraph 0056 for a neural network with an “attention-based transformed architecture” including spatial and temporal encoder blocks. This means there can be a spatial block and a temporal block. See paragraph 0057 for multi-axes attention blocks that “compute attention scores”. Each attention block can including “one or more multi-head self-attention layers”. These blocks can be stacked together. See also Fig. 4. See also paragraph 0053 for a “temporal dimension”.) and
a second attention head configured to concurrently apply second self-attention to the plurality of input feature vectors across the feature dimension of the latent space (see Yan paragraph 0056 for a neural network with an “attention-based transformed architecture” including spatial and temporal encoder blocks. This means there can be a spatial block and a temporal block. See paragraph 0057 for multi-axes attention blocks that “compute attention scores”. Each attention block can including “one or more multi-head self-attention layers”. These blocks can be stacked together. See also Fig. 4. See paragraph 0085 for the operations being performed “concurrently”).
Regarding claim 3, Yan discloses the method of Claim 1.
Yan further discloses:
The method of Claim 1, wherein
the Al based system operates with a computer implemented traffic assistant to cause one or more vehicle propulsion operations to be controlled based at least in part on one or more of
vehicle velocity predictions (see paragraph 0031 for “predicted locations/velocities”. See paragraph 0088 for probable trajectories of the vehicle including “velocities of the vehicle”.) or
lane changes for the vehicle in the one or more target predictions (see paragraph 0031 for “predicted locations/velocities”. See paragraph 0088 for probable trajectories of the vehicle including “velocities of the vehicle”. See paragraph 0088 for “the acceleration/braking status, and “steering status, status of signaling” etc of the vehicle. See paragraph 0093 for determining that one predicted trajectory has a low probability of being completed, and therefore the system can “change driving lane” or perform some other “selected driving path”.).
Regarding claim 4, Yan discloses the method of Claim 1.
Yan further discloses:
The method of Claim 1, wherein
the sensor-acquired data is generated with a sensor stack deployed with the vehicle (see paragraph 0014);
wherein the sensor stack includes one or more of: in-vehicle cameras, image sensors, non-image sensors, radars, LIDARs, ultrasonic sensors, or infrared sensors (see paragraph 0014).
Regarding claim 5, Yan discloses the method of Claim 1.
Yan further discloses:
The method of Claim 1, wherein
the non-sensor-acquired data includes one or more lane images (see paragraph 0018 for input data including “roadgraph data (e.g., map data, lane boundaries, road signs, etc.)”. See also paragraph 0044, especially the first sentence. See also paragraph 0045 for a data repository 250 with roadgraphs and “images”.).
Regarding claim 6, Yan discloses the method of Claim 1.
Yan further discloses:
The method of Claim 1, wherein
the one or more target predictions are generated by one or more multi-layer perceptron neural networks based at least in part on the attention scores forming the plurality of attention layers as generated by the plurality of attention heads of the one or more transformer neural networks (see Yan paragraph 0056 for a neural network with an “attention-based transformed architecture” including spatial and temporal encoder blocks. See paragraph 0057 for multi-axes attention blocks that “compute attention scores”. Each attention block can including “one or more multi-head self-attention layers”. These blocks can be stacked together. See also Fig. 4. See paragraph 0057 for a “multilayer perception layers” that are “elements” of the neural network”).
Regarding claim 7, Yan discloses the method of Claim 1.
Yan further discloses:
The method of Claim 1, wherein
one or both of a car network graph in reference to the vehicle (see paragraph 0018) or a trajectory of the vehicle is generated based at least in part on the one or more target predictions (see paragraph 0093 for determining that one predicted trajectory has a low probability of being completed, and therefore the system can “change driving lane” or perform some other “selected driving path”. See paragraph 0017 for the model being able to determine who other road users will respond to a host vehicle’s driving path.).
Regarding claim 8, Yan discloses:
A system, comprising (see Figs. 1 and 2):
one or more computing processors (see Fig. 7, item 702);
one or more non-transitory computer readable media storing a program of instructions that is executable by the one or more computing processors to perform (see Fig. 7, item 704 and 718):
determining, from a plurality of physical object observations derived from sensor-acquired data and non-sensor-acquired data collected for a vehicle, a plurality of input feature vectors in a latent space of three dimensions (see Yan paragraph 0014 for the sensors on the host vehicle. See Fig. 2 for sensing data 210, traffic lights 220, and map information 124. See paragraph 0032 for additional sensors. See paragraph 0052 for “feature vectors” in “different dimensions”. See paragraph 0054 for the “feature vector” including in “a multi-dimensional embedding space”. See paragraph 0058 for tokens in three dimensions.),
wherein the three dimensions include
a time dimension (see paragraph 0053 for a “temporal dimension”),
a feature dimension and (the present published disclosure, Shamsoshoara (US2025/0196892) uses the word “feature” extremely broadly. Paragraph 0125 refers to “different feature types (e.g., different vehicles/objects)”. With that in mind, see Yan, paragraph 0052 for “feature vectors” such as roadgraph information having a dimension Drx1.)
an embedding size dimension (in the present published disclosure see paragraph 0101 for the “The embeddings or vectors may be of an equal length (or size/dimension) referred to as ‘embedding size’”. With that in mind, see Yan paragraph 0052 for “embeddings of a common dimension”.);
applying a plurality of attention heads implemented in one or more transformer networks of an artificial intelligence (AI)-based to the input feature vectors in the latent space to generate attention scores forming a plurality of attention layers (see Yan paragraph 0056 for a neural network with an “attention-based transformed architecture” including spatial and temporal encoder blocks. See paragraph 0057 for multi-axes attention blocks that “compute attention scores”. Each attention block can including “one or more multi-head self-attention layers”. These blocks can be stacked together. See also Fig. 4);
generating, based at least in part on the attention scores in the plurality of attention layers, one or more target predictions relating to navigation operations of the vehicle (see paragraph 0033 for the system being configured for “selecting a particular path through the immediate driving environment, which can include selecting a traffic lane, negotiating a traffic congestion…and so on.” See Fig. 4 and paragraph 0071 for “predicted trajectories 442”.);
interacting with a driver assistant sub-system of the vehicle to cause a vehicle propulsion operation to be performed in accordance with the one or more target predictions (see Yan paragraph 0033 for the system being configured for “selecting a particular path through the immediate driving environment, which can include selecting a traffic lane, negotiating a traffic congestion…and so on.” See also paragraph 0017.).
Regarding claims 9-14, they are substantially similar to claims 2-7, respectively. Please see the rejections of those claims.
Regarding claim 22, Yan discloses:
One or more non-transitory computer readable media storing a program of instructions that is executable by one or more computing processors to perform (see Fig. 7, items 704 and 718):
determining, from a plurality of physical object observations derived from sensor- acquired data and non-sensor-acquired data collected for a vehicle, a plurality of input feature vectors in a latent space of three dimensions (see Yan paragraph 0014 for the sensors on the host vehicle. See Fig. 2 for sensing data 210, traffic lights 220, and map information 124. See paragraph 0032 for additional sensors. See paragraph 0052 for “feature vectors” in “different dimensions”. See paragraph 0054 for the “feature vector” including in “a multi-dimensional embedding space”. See paragraph 0058 for tokens in three dimensions.),
wherein the three dimensions include
a time dimension (see paragraph 0053 for a “temporal dimension”),
a feature dimension and (the present published disclosure, Shamsoshoara (US2025/0196892) uses the word “feature” extremely broadly. Paragraph 0125 refers to “different feature types (e.g., different vehicles/objects)”. With that in mind, see Yan, paragraph 0052 for “feature vectors” such as roadgraph information having a dimension Drx1.)
an embedding size dimension (in the present published disclosure see paragraph 0101 for the “The embeddings or vectors may be of an equal length (or size/dimension) referred to as ‘embedding size’”. With that in mind, see Yan paragraph 0052 for “embeddings of a common dimension”.);
applying a plurality of attention heads implemented in one or more transformer networks of an artificial intelligence (AI) based system to the input feature vectors in the latent space to generate attention scores forming a plurality of attention layers (see Yan paragraph 0056 for a neural network with an “attention-based transformed architecture” including spatial and temporal encoder blocks. See paragraph 0057 for multi-axes attention blocks that “compute attention scores”. Each attention block can including “one or more multi-head self-attention layers”. These blocks can be stacked together. See also Fig. 4);
generating, based at least in part on the attention scores in the plurality of attention layers, one or more target predictions relating to navigation operations of the vehicle (see paragraph 0033 for the system being configured for “selecting a particular path through the immediate driving environment, which can include selecting a traffic lane, negotiating a traffic congestion…and so on.” See Fig. 4 and paragraph 0071 for “predicted trajectories 442”.);
interacting with a driver assistant sub-system of the vehicle to cause a vehicle propulsion operation to be performed in accordance with the one or more target predictions (see Yan paragraph 0033 for the system being configured for “selecting a particular path through the immediate driving environment, which can include selecting a traffic lane, negotiating a traffic congestion…and so on.” See also paragraph 0017.).
Regarding claims 23-28, they are substantially similar to claims 2-7, respectively. Please see the rejections of those claims.
Additional Art
The prior art made of record here, though not relied upon, is considered pertinent to the present disclosure.
One close prior art is Wang et al. (U.S. 11,774,505). Wang teaches in at least claim 1 and Fig. 2 a deep learning system that can input a third matrix into a “series-connected and trained multi-head self-attention layer” in order to identify battery safety hazards. “The trained multi-head self-attention layer performs global correlation analysis on the third matrix”. According to the disclosure the self-attention layer “can extract the correlation between any two information dimensions, which is beneficial to elimination of the interference of an arrangement order of different information dimensions to the extraction of global features.”
Another close prior art is Hatamizadeh et al. (US2023/0145535). This teaches “multi-headed self-attention sub-layers” in at least paragraph 0073. See paragraph 0147 for a system that receives information about vehicle actions such as lane changes that a vehicle is making or will make. See at least claim 3 for a neural network that encodes features of a first and second dataset to a shared latent space. See Fig. 7, paragraph 0094 and claim 4 for the first dataset relating to image data and the second dataset comprising textual descriptions of the corresponding image data in the first dataset. Does not teach an “attention score.”
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DANIEL M. ROBERT whose telephone number is (571)270-5841. The examiner can normally be reached M-F 7:30-4:30 EST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Hunter Lonsberry can be reached at 571-272-7298. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/DANIEL M. ROBERT/Primary Examiner, Art Unit 3665